population, because each individual should be assigned a meaningful score. However, according to Herrmann (1969) and Guilford (1959), a dimension may not necessarily be general or universal for the total population. More specifically, Cattell (1965) argues that it is not appropriate to test the dimensionality of highly specific characteristics in samples where the majority of members are not affected by the property in question. (3) Even in the RS there are still criteria with very low prevalence, including “legal problems” (1.7%) or “neglect roles” (3.7%). In item analysis these items would usually be excluded. Although only individuals with at least one endorsed criterion were included in the RS, these items may still inflate homogeneity. (4) Selection bias may have occurred; specifically, heavy drinkers with high probability of reporting AUD criteria may be underrepresented.